mirror of
https://github.com/facebookresearch/ReAgent.git
synced 2026-05-17 12:40:39 +00:00
Zhengxing Chen
169 lines
9.1 KiB
HTML
169 lines
9.1 KiB
HTML
<!DOCTYPE html>
|
||
<html class="writer-html5" lang="en" >
|
||
<head>
|
||
<meta charset="utf-8" /><meta name="generator" content="Docutils 0.17.1: http://docutils.sourceforge.net/" />
|
||
|
||
<meta name="viewport" content="width=device-width, initial-scale=1.0" />
|
||
<title>Training a model across multiple GPUs — ReAgent 1.0 documentation</title>
|
||
<link rel="stylesheet" href="_static/pygments.css" type="text/css" />
|
||
<link rel="stylesheet" href="_static/css/theme.css" type="text/css" />
|
||
<!--[if lt IE 9]>
|
||
<script src="_static/js/html5shiv.min.js"></script>
|
||
<![endif]-->
|
||
|
||
<script data-url_root="./" id="documentation_options" src="_static/documentation_options.js"></script>
|
||
<script src="_static/jquery.js"></script>
|
||
<script src="_static/underscore.js"></script>
|
||
<script src="_static/doctools.js"></script>
|
||
<script src="_static/js/theme.js"></script>
|
||
<link rel="index" title="Index" href="genindex.html" />
|
||
<link rel="search" title="Search" href="search.html" />
|
||
</head>
|
||
|
||
<body class="wy-body-for-nav">
|
||
<div class="wy-grid-for-nav">
|
||
<nav data-toggle="wy-nav-shift" class="wy-nav-side">
|
||
<div class="wy-side-scroll">
|
||
<div class="wy-side-nav-search" >
|
||
<a href="index.html" class="icon icon-home"> ReAgent
|
||
</a>
|
||
<div role="search">
|
||
<form id="rtd-search-form" class="wy-form" action="search.html" method="get">
|
||
<input type="text" name="q" placeholder="Search docs" />
|
||
<input type="hidden" name="check_keywords" value="yes" />
|
||
<input type="hidden" name="area" value="default" />
|
||
</form>
|
||
</div>
|
||
</div><div class="wy-menu wy-menu-vertical" data-spy="affix" role="navigation" aria-label="Navigation menu">
|
||
<p class="caption" role="heading"><span class="caption-text">Getting Started</span></p>
|
||
<ul>
|
||
<li class="toctree-l1"><a class="reference internal" href="installation.html">Installation</a></li>
|
||
<li class="toctree-l1"><a class="reference internal" href="usage.html">Usage</a></li>
|
||
<li class="toctree-l1"><a class="reference internal" href="rasp_tutorial.html">RASP (Not Actively Maintained)</a></li>
|
||
</ul>
|
||
<p class="caption" role="heading"><span class="caption-text">Advanced Topics</span></p>
|
||
<ul>
|
||
<li class="toctree-l1"><a class="reference internal" href="continuous_integration.html">Continuous Integration</a></li>
|
||
</ul>
|
||
<p class="caption" role="heading"><span class="caption-text">Package Reference</span></p>
|
||
<ul>
|
||
<li class="toctree-l1"><a class="reference internal" href="api/reagent.core.html">Core</a></li>
|
||
<li class="toctree-l1"><a class="reference internal" href="api/reagent.data.html">Data</a></li>
|
||
<li class="toctree-l1"><a class="reference internal" href="api/reagent.gym.html">Gym</a></li>
|
||
<li class="toctree-l1"><a class="reference internal" href="api/reagent.evaluation.html">Evaluation</a></li>
|
||
<li class="toctree-l1"><a class="reference internal" href="api/reagent.lite.html">Lite</a></li>
|
||
<li class="toctree-l1"><a class="reference internal" href="api/reagent.mab.html">MAB</a></li>
|
||
<li class="toctree-l1"><a class="reference internal" href="api/reagent.model_managers.html">Model Managers</a></li>
|
||
<li class="toctree-l1"><a class="reference internal" href="api/reagent.model_utils.html">Model Utils</a></li>
|
||
<li class="toctree-l1"><a class="reference internal" href="api/reagent.net_builder.html">Net Builders</a></li>
|
||
<li class="toctree-l1"><a class="reference internal" href="api/reagent.optimizer.html">Optimizers</a></li>
|
||
<li class="toctree-l1"><a class="reference internal" href="api/reagent.models.html">Models</a></li>
|
||
<li class="toctree-l1"><a class="reference internal" href="api/reagent.prediction.html">Prediction</a></li>
|
||
<li class="toctree-l1"><a class="reference internal" href="api/reagent.preprocessing.html">Preprocessing</a></li>
|
||
<li class="toctree-l1"><a class="reference internal" href="api/reagent.training.html">Training</a></li>
|
||
<li class="toctree-l1"><a class="reference internal" href="api/reagent.workflow.html">Workflow</a></li>
|
||
<li class="toctree-l1"><a class="reference internal" href="api/modules.html">All Modules</a></li>
|
||
</ul>
|
||
<p class="caption" role="heading"><span class="caption-text">Others</span></p>
|
||
<ul>
|
||
<li class="toctree-l1"><a class="reference external" href="https://github.com/facebookresearch/ReAgent">Github</a></li>
|
||
<li class="toctree-l1"><a class="reference internal" href="license.html">License</a></li>
|
||
</ul>
|
||
|
||
</div>
|
||
</div>
|
||
</nav>
|
||
|
||
<section data-toggle="wy-nav-shift" class="wy-nav-content-wrap"><nav class="wy-nav-top" aria-label="Mobile navigation menu" >
|
||
<i data-toggle="wy-nav-top" class="fa fa-bars"></i>
|
||
<a href="index.html">ReAgent</a>
|
||
</nav>
|
||
|
||
<div class="wy-nav-content">
|
||
<div class="rst-content">
|
||
<div role="navigation" aria-label="Page navigation">
|
||
<ul class="wy-breadcrumbs">
|
||
<li><a href="index.html" class="icon icon-home"></a> »</li>
|
||
<li>Training a model across multiple GPUs</li>
|
||
<li class="wy-breadcrumbs-aside">
|
||
<a href="_sources/distributed.rst.txt" rel="nofollow"> View page source</a>
|
||
</li>
|
||
</ul>
|
||
<hr/>
|
||
</div>
|
||
<div role="main" class="document" itemscope="itemscope" itemtype="http://schema.org/Article">
|
||
<div itemprop="articleBody">
|
||
|
||
<section id="training-a-model-across-multiple-gpus">
|
||
<span id="distributed"></span><h1>Training a model across multiple GPUs<a class="headerlink" href="#training-a-model-across-multiple-gpus" title="Permalink to this headline"></a></h1>
|
||
<p>Before we get started, please check out the <a class="reference internal" href="usage.html#usage"><span class="std std-ref">Usage Guide</span></a> and
|
||
the <a class="reference external" href="https://pytorch.org/docs/stable/distributed.html">PyTorch Distributed documentation</a>.</p>
|
||
<section id="how-distributed-training-works">
|
||
<h2>How distributed training works<a class="headerlink" href="#how-distributed-training-works" title="Permalink to this headline"></a></h2>
|
||
<p>With a single GPU and model, training follows this process:</p>
|
||
<ol class="arabic simple">
|
||
<li><p>Compute the loss from a minibatch of data (the forward pass of the model)</p></li>
|
||
<li><p>Backpropagate that loss through the model to compute gradients (the backward pass of the model)</p></li>
|
||
<li><p>Sum the gradients across the minibatch.</p></li>
|
||
<li><p>Run the optimizer by calling the “step()” function.</p></li>
|
||
</ol>
|
||
<p>Now assume we have several GPUs, and they each have exactly the same model.
|
||
#1 and #2 are <a class="reference external" href="https://en.wikipedia.org/wiki/Embarrassingly_parallel">embarrassingly parallel</a> and can be distributed to many nodes.
|
||
As long as we can sum across nodes to complete #3 (this is known as an ‘all-reduce’), then each node can run #4 on the same gradients,
|
||
and the resulting models will again be identical. This is the premise behind distributed training.</p>
|
||
</section>
|
||
<section id="training-on-a-single-node">
|
||
<h2>Training on a single node<a class="headerlink" href="#training-on-a-single-node" title="Permalink to this headline"></a></h2>
|
||
<p>Using multiple GPUs on a single node is relatively straightforward. When running either the dqn_workflow or the parametric_dqn_workflow,
|
||
set the “use_all_avail_gpus” parameters in the input config (the json file) to true. ReAgent will detect the number of available GPUs and
|
||
run on all of them without any additional effort.</p>
|
||
</section>
|
||
<section id="training-on-multiple-nodes">
|
||
<h2>Training on multiple nodes<a class="headerlink" href="#training-on-multiple-nodes" title="Permalink to this headline"></a></h2>
|
||
<p>Multi-node training requires more setup. Some prerequisites:</p>
|
||
<ol class="arabic simple">
|
||
<li><p>A networked filesystem (such as NFS) that all trainers can access</p></li>
|
||
<li><p>A local filesystem that is unique to each trainer</p></li>
|
||
</ol>
|
||
<p>To set up the environment for <em>N</em> nodes, do the following:</p>
|
||
<ol class="arabic simple">
|
||
<li><p>Create an empty file on the networked filesystem which the <em>N</em> machines will use to communicate</p></li>
|
||
<li><p>Either split the dataset into <em>N</em> equal parts, <strong>or</strong> shuffle the dataset to create N copies.</p></li>
|
||
<li><p>Put one copy of the dataset onto each machine’s local filesystem in exactly the same path.</p></li>
|
||
<li><p>Set the “use_all_avail_gpus” parameter to true as above</p></li>
|
||
<li><p>Also set the “num_nodes” parameter to <em>N</em>.</p></li>
|
||
<li><p>On each machine, run the workflow with the “–node_index=n” flag, where n is the index of that machine.</p></li>
|
||
<li><p>The machine with –node_index=0 will save the final model to the output path specified.</p></li>
|
||
</ol>
|
||
</section>
|
||
</section>
|
||
|
||
|
||
</div>
|
||
</div>
|
||
<footer>
|
||
|
||
<hr/>
|
||
|
||
<div role="contentinfo">
|
||
<p>© Copyright 2022, Meta Platforms, Inc.</p>
|
||
</div>
|
||
|
||
Built with <a href="https://www.sphinx-doc.org/">Sphinx</a> using a
|
||
<a href="https://github.com/readthedocs/sphinx_rtd_theme">theme</a>
|
||
provided by <a href="https://readthedocs.org">Read the Docs</a>.
|
||
|
||
|
||
</footer>
|
||
</div>
|
||
</div>
|
||
</section>
|
||
</div>
|
||
<script>
|
||
jQuery(function () {
|
||
SphinxRtdTheme.Navigation.enable(true);
|
||
});
|
||
</script>
|
||
|
||
</body>
|
||
</html> |